organoid_tracker.neural_network.position_detection_cnn.training_data_creator module

class organoid_tracker.neural_network.position_detection_cnn.training_data_creator.ImageWithPositions(experiment_name: str, images: Images, time_point: TimePoint, xyz_positions: ndarray)

Bases: object

create_labels(image_size_zyx: Tuple[int, int, int], *, image_offset_zyx: Tuple[int, int, int] = (0, 0, 0))

Creates an image with the number 1 at self.xyz_positions. Ignores positions outside the image. Allows you to specify an offset and size, which makes it possible to draw the labels for any crop.

experiment_name: str

get_image_offset() → Position

Gets the offset of the image.

get_image_size_zyx(dt: int = 0) → Tuple[int, int, int]

Gets the shape of the image. This method tries to avoid loading the actual image data to find it out.

load_image(dt: int = 0) → Optional[ndarray]

load_image_time_stack(time_window: Union[List[int], Tuple[int, int]] = (0, 0), delay=0) → Optional[ndarray]

Loads images in a time window. Returns a 4D array, [z, y, x, t].

time_point: TimePoint

xyz_positions: ndarray

organoid_tracker.neural_network.position_detection_cnn.training_data_creator.create_image_list_without_positions(experiment: Experiment) → List[ImageWithPositions]

organoid_tracker.neural_network.position_detection_cnn.training_data_creator.create_image_with_positions_list(experiments: Iterable[Experiment])